@delali/narsil 0.1.0

package/dist/languages/twi.mjs

@@ -0,0 +1,84 @@
+// src/languages/twi.ts
+var stopWords = /* @__PURE__ */ new Set([
+  "me",
+  "wo",
+  "\u0254no",
+  "\u025Bno",
+  "y\u025Bn",
+  "mo",
+  "w\u0254n",
+  "no",
+  "ne",
+  "na",
+  "nanso",
+  "nti",
+  "anaa",
+  "ansa",
+  "s\u025B",
+  "as\u025B",
+  "\u025Bfiri",
+  "\u025Bsiane",
+  "s\u025Bde\u025B",
+  "mpo",
+  "ky\u025Bn",
+  "a",
+  "yi",
+  "de",
+  "y\u025B",
+  "nso",
+  "ma",
+  "fi",
+  "k\u0254",
+  "ba",
+  "ka",
+  "bi",
+  "mu",
+  "so",
+  "ho",
+  "ase",
+  "fa",
+  "hwe",
+  "b\u025B",
+  "re",
+  "ara",
+  "paa",
+  "kakra",
+  "wei",
+  "eyi",
+  "eno",
+  "emu",
+  "eso",
+  "eho",
+  "nyinaa",
+  "obi",
+  "obiara",
+  "hwee",
+  "ankasa",
+  "nkoraa",
+  "te",
+  "san",
+  "kyer\u025B",
+  "da",
+  "akyire",
+  "bere",
+  "se",
+  "enti",
+  "oo",
+  "mmom",
+  "kor\u0254",
+  "gye",
+  "nye",
+  "nnye"
+]);
+var twi = {
+  name: "twi",
+  stemmer: null,
+  stopWords,
+  tokenizer: {
+    splitPattern: /[^a-zA-ZɛɔƐƆ0-9]+/gi,
+    normalizeDiacritics: false,
+    minTokenLength: 1
+  }
+};
+
+export { twi };

package/dist/languages/ukrainian.d.ts

@@ -0,0 +1,5 @@
+import { L as LanguageModule } from '../language-ClUGLuoC.js';
+
+declare const ukrainian: LanguageModule;
+
+export { ukrainian };

package/dist/languages/ukrainian.mjs

@@ -0,0 +1,444 @@
+// src/languages/ukrainian.ts
+var VOWELS = "\u0430\u0435\u0438\u043E\u0443\u044E\u044F\u0456\u0457\u0454";
+function isVowel(ch) {
+  return VOWELS.includes(ch);
+}
+function findRegions(word) {
+  let rv = word.length;
+  let r1 = word.length;
+  let r2 = word.length;
+  for (let i = 0; i < word.length; i++) {
+    if (isVowel(word[i])) {
+      rv = i + 1;
+      break;
+    }
+  }
+  for (let i = 1; i < word.length; i++) {
+    if (!isVowel(word[i]) && isVowel(word[i - 1])) {
+      r1 = i + 1;
+      break;
+    }
+  }
+  for (let i = r1 + 1; i < word.length; i++) {
+    if (!isVowel(word[i]) && isVowel(word[i - 1])) {
+      r2 = i + 1;
+      break;
+    }
+  }
+  return { rv, r1, r2 };
+}
+function removeSuffix(word, suffix, regionStart) {
+  if (!word.endsWith(suffix)) return null;
+  if (word.length - suffix.length < regionStart) return null;
+  return word.slice(0, word.length - suffix.length);
+}
+function findAndRemoveSuffix(word, suffixes, regionStart) {
+  for (const suffix of suffixes) {
+    const result = removeSuffix(word, suffix, regionStart);
+    if (result !== null) return result;
+  }
+  return null;
+}
+var PERFECTIVE_GERUND_GROUP1 = ["\u0432\u0448\u0438\u0441\u044C", "\u0432\u0448\u0438", "\u0432"];
+var PERFECTIVE_GERUND_GROUP2 = ["\u0456\u0432\u0448\u0438\u0441\u044C", "\u0438\u0432\u0448\u0438", "\u0456\u0432"];
+function removePerfectiveGerund(word, rv) {
+  for (const suffix of PERFECTIVE_GERUND_GROUP1) {
+    if (word.endsWith(suffix) && word.length - suffix.length >= rv) {
+      const base = word.slice(0, word.length - suffix.length);
+      if (base.endsWith("\u0430") || base.endsWith("\u044F")) {
+        return base;
+      }
+    }
+  }
+  return findAndRemoveSuffix(word, PERFECTIVE_GERUND_GROUP2, rv);
+}
+var REFLEXIVE_SUFFIXES = ["\u0441\u044F", "\u0441\u044C"];
+function removeReflexive(word, rv) {
+  return findAndRemoveSuffix(word, REFLEXIVE_SUFFIXES, rv);
+}
+var ADJECTIVE_SUFFIXES = [
+  "\u0438\u043C\u0438",
+  "\u043E\u0433\u043E",
+  "\u043E\u043C\u0443",
+  "\u0456\u0439",
+  "\u0438\u0439",
+  "\u0457\u0439",
+  "\u043E\u044E",
+  "\u0435\u0435",
+  "\u0456\u0435",
+  "\u0456\u0454",
+  "\u0438\u0435",
+  "\u043E\u0457",
+  "\u0457\u0439",
+  "\u0435\u043C",
+  "\u0456\u043C",
+  "\u043E\u043C",
+  "\u0438\u043C",
+  "\u0438\u0445",
+  "\u0456\u0445",
+  "\u0443\u044E",
+  "\u044E\u044E",
+  "\u0430\u044F",
+  "\u044F\u044F",
+  "\u043E\u0435",
+  "\u0435\u0454"
+];
+var PARTICIPLE_GROUP1 = ["\u0435\u043C", "\u043D\u043D", "\u0432\u0448", "\u044E\u0449", "\u0449"];
+var PARTICIPLE_GROUP2 = ["\u0456\u0432\u0448", "\u0438\u0432\u0448", "\u0443\u044E\u0447", "\u0443\u044E\u0449"];
+function removeAdjectival(word, rv) {
+  const adjResult = findAndRemoveSuffix(word, ADJECTIVE_SUFFIXES, rv);
+  if (adjResult === null) return null;
+  for (const suffix of PARTICIPLE_GROUP1) {
+    if (adjResult.endsWith(suffix) && adjResult.length - suffix.length >= rv) {
+      const base = adjResult.slice(0, adjResult.length - suffix.length);
+      if (base.endsWith("\u0430") || base.endsWith("\u044F")) {
+        return base;
+      }
+    }
+  }
+  const partResult = findAndRemoveSuffix(adjResult, PARTICIPLE_GROUP2, rv);
+  if (partResult !== null) return partResult;
+  return adjResult;
+}
+var VERB_GROUP1 = [
+  "\u0435\u0442\u0435",
+  "\u0439\u0442\u0435",
+  "\u0435\u0448\u044C",
+  "\u043D\u043D\u043E",
+  "\u043B\u0430",
+  "\u043D\u0430",
+  "\u043B\u0438",
+  "\u0435\u043C",
+  "\u043B\u043E",
+  "\u043D\u043E",
+  "\u0435\u0442",
+  "\u044E\u0442",
+  "\u043D\u0438",
+  "\u0442\u044C",
+  "\u0439",
+  "\u043B",
+  "\u043D"
+];
+var VERB_GROUP2 = [
+  "\u0435\u0439\u0442\u0435",
+  "\u0443\u0439\u0442\u0435",
+  "\u0443\u044E\u0442\u044C",
+  "\u0438\u043B\u0430",
+  "\u0438\u043B\u0430",
+  "\u0435\u043D\u0430",
+  "\u0456\u0442\u0435",
+  "\u0438\u0442\u0435",
+  "\u0438\u043B\u0438",
+  "\u0438\u043B\u043E",
+  "\u0435\u043D\u043E",
+  "\u044F\u0442\u044C",
+  "\u0438\u0442\u044C",
+  "\u0456\u0442\u044C",
+  "\u0456\u0448\u044C",
+  "\u0435\u0439",
+  "\u0443\u0439",
+  "\u0438\u0432",
+  "\u0438\u043B",
+  "\u0456\u043C",
+  "\u0438\u043C",
+  "\u0435\u043D",
+  "\u0443\u0454",
+  "\u0443\u044E",
+  "\u044E"
+];
+function removeVerb(word, rv) {
+  for (const suffix of VERB_GROUP1) {
+    if (word.endsWith(suffix) && word.length - suffix.length >= rv) {
+      const base = word.slice(0, word.length - suffix.length);
+      if (base.endsWith("\u0430") || base.endsWith("\u044F")) {
+        return base;
+      }
+    }
+  }
+  return findAndRemoveSuffix(word, VERB_GROUP2, rv);
+}
+var NOUN_SUFFIXES = [
+  "\u0456\u044F\u043C\u0438",
+  "\u044F\u043C\u0438",
+  "\u0456\u044F\u043C",
+  "\u0456\u0454\u043C",
+  "\u0456\u0454\u044E",
+  "\u0430\u043C\u0438",
+  "\u0435\u0432\u0456",
+  "\u043E\u0432\u0456",
+  "\u0456\u0432",
+  "\u0457\u0432",
+  "\u0456\u0439",
+  "\u0435\u0439",
+  "\u0435\u043C",
+  "\u0454\u043C",
+  "\u0454\u044E",
+  "\u043E\u0432",
+  "\u043E\u0439",
+  "\u043E\u043C",
+  "\u043E\u044E",
+  "\u0456\u044E",
+  "\u044C\u044E",
+  "\u0456\u044F",
+  "\u044C\u044F",
+  "\u044F\u043C",
+  "\u044F\u0445",
+  "\u0430\u0445",
+  "\u0430\u043C",
+  "\u0430",
+  "\u0435",
+  "\u0454",
+  "\u0456",
+  "\u0457",
+  "\u0438",
+  "\u0439",
+  "\u043E",
+  "\u0443",
+  "\u044C",
+  "\u044E",
+  "\u044F"
+];
+function removeNoun(word, rv) {
+  return findAndRemoveSuffix(word, NOUN_SUFFIXES, rv);
+}
+var DERIVATIONAL_SUFFIXES = ["\u0456\u0441\u0442\u044C", "\u043E\u0441\u0442\u044C", "\u0456\u0441\u0442", "\u043E\u0441\u0442"];
+function removeDerivational(word, r2) {
+  const result = findAndRemoveSuffix(word, DERIVATIONAL_SUFFIXES, r2);
+  return result ?? word;
+}
+function step1(word, rv) {
+  const pgResult = removePerfectiveGerund(word, rv);
+  if (pgResult !== null) return pgResult;
+  const reflexResult = removeReflexive(word, rv);
+  const base = reflexResult ?? word;
+  const adjResult = removeAdjectival(base, rv);
+  if (adjResult !== null) return adjResult;
+  const verbResult = removeVerb(base, rv);
+  if (verbResult !== null) return verbResult;
+  const nounResult = removeNoun(base, rv);
+  if (nounResult !== null) return nounResult;
+  return base;
+}
+function step2(word, rv) {
+  if (word.endsWith("\u0438") && word.length - 1 >= rv) {
+    return word.slice(0, -1);
+  }
+  return word;
+}
+function step3(word, r2) {
+  return removeDerivational(word, r2);
+}
+function step4(word, rv) {
+  if (word.endsWith("\u0435\u0439\u0448\u0435") && word.length - 4 >= rv) {
+    return word.slice(0, -4);
+  }
+  if (word.endsWith("\u0435\u0439\u0448") && word.length - 3 >= rv) {
+    return word.slice(0, -3);
+  }
+  if (word.endsWith("\u043D\u043D") && word.length - 2 >= rv) {
+    return word.slice(0, -1);
+  }
+  if (word.endsWith("\u044C") && word.length - 1 >= rv) {
+    return word.slice(0, -1);
+  }
+  return word;
+}
+function stem(word) {
+  const { rv, r2 } = findRegions(word);
+  word = step1(word, rv);
+  word = step2(word, rv);
+  word = step3(word, r2);
+  word = step4(word, rv);
+  return word;
+}
+var stopWords = /* @__PURE__ */ new Set([
+  "\u0430",
+  "\u0430\u043B\u0435",
+  "\u0431",
+  "\u0431\u0435\u0437",
+  "\u0431\u0438",
+  "\u0431\u0456\u043B\u044F",
+  "\u0431\u0456\u043B\u044C\u0448",
+  "\u0431\u043E",
+  "\u0431\u0443\u0432",
+  "\u0431\u0443\u043B\u0430",
+  "\u0431\u0443\u043B\u0438",
+  "\u0431\u0443\u043B\u043E",
+  "\u0431\u0443\u0442\u0438",
+  "\u0432",
+  "\u0432\u0430\u043C",
+  "\u0432\u0430\u0441",
+  "\u0432\u0430\u0448",
+  "\u0432\u0430\u0448\u0430",
+  "\u0432\u0430\u0448\u0435",
+  "\u0432\u0430\u0448\u0456",
+  "\u0432\u0435\u0441\u044C",
+  "\u0432\u0438",
+  "\u0432\u0456\u0434",
+  "\u0432\u0456\u043D",
+  "\u0432\u043E\u043D\u0430",
+  "\u0432\u043E\u043D\u043E",
+  "\u0432\u043E\u043D\u0438",
+  "\u0432\u0441\u0435",
+  "\u0432\u0441\u044F",
+  "\u0432\u0441\u0456",
+  "\u0432\u0441\u044E",
+  "\u0433",
+  "\u0434",
+  "\u0434\u0430",
+  "\u0434\u0430\u043B\u0438",
+  "\u0434\u0432\u0430",
+  "\u0434\u0435",
+  "\u0434\u0435\u0449\u043E",
+  "\u0434\u043E",
+  "\u0434\u043E\u0431\u0440\u0435",
+  "\u0434\u0440\u0443\u0433\u0438\u0439",
+  "\u0435",
+  "\u0454",
+  "\u0436",
+  "\u0437",
+  "\u0437\u0430",
+  "\u0437\u0430\u0440\u0430\u0437",
+  "\u0437\u0456",
+  "\u0456",
+  "\u0456\u0437",
+  "\u0456\u0439",
+  "\u0456\u043D\u043A\u043E\u043B\u0438",
+  "\u0456\u043D\u0448\u0438\u0439",
+  "\u0457\u0439",
+  "\u0457\u043C",
+  "\u0457\u0445",
+  "\u0457\u0457",
+  "\u0439",
+  "\u0439\u043E\u0433\u043E",
+  "\u0439\u043E\u043C\u0443",
+  "\u043A",
+  "\u043A\u043E\u0436\u0435\u043D",
+  "\u043A\u043E\u0436\u043D\u0430",
+  "\u043A\u043E\u043B\u0438",
+  "\u043A\u043E\u0442\u0440\u0438\u0439",
+  "\u043A\u043E\u0442\u0440\u0430",
+  "\u043A\u043E\u043C\u0443",
+  "\u043A\u0440\u0456\u043C",
+  "\u043B",
+  "\u043C",
+  "\u043C\u0438",
+  "\u043C\u0456\u0436",
+  "\u043C\u0456\u0439",
+  "\u043C\u043D\u043E\u044E",
+  "\u043C\u043E\u0433\u043E",
+  "\u043C\u043E\u0436\u0435",
+  "\u043C\u043E\u0436\u043D\u0430",
+  "\u043C\u043E\u044F",
+  "\u043C\u043E\u0454",
+  "\u043C\u043E\u0454\u043C\u0443",
+  "\u043C\u043E\u0457\u0439",
+  "\u043C\u043E\u0457\u0445",
+  "\u043C\u043E\u044E",
+  "\u043D",
+  "\u043D\u0430",
+  "\u043D\u0430\u0432\u0456\u0442\u044C",
+  "\u043D\u0430\u0434",
+  "\u043D\u0430\u043C",
+  "\u043D\u0430\u0441",
+  "\u043D\u0430\u0448",
+  "\u043D\u0430\u0448\u0430",
+  "\u043D\u0430\u0448\u0435",
+  "\u043D\u0430\u0448\u0438\u0445",
+  "\u043D\u0435",
+  "\u043D\u0456",
+  "\u043D\u0456\u0436",
+  "\u043D\u0456\u043A\u043E\u043B\u0438",
+  "\u043D\u0456\u0447\u043E\u0433\u043E",
+  "\u043E",
+  "\u043E\u0431\u0438\u0434\u0432\u0430",
+  "\u043E\u0434\u0438\u043D",
+  "\u043E\u0434\u043D\u0430\u043A",
+  "\u043E\u0434\u043D\u0430",
+  "\u043E\u0434\u043D\u0435",
+  "\u043E\u0434\u043D\u0456",
+  "\u043F",
+  "\u043F\u0456\u0434",
+  "\u043F\u0456\u0441\u043B\u044F",
+  "\u043F\u043E",
+  "\u043F\u0440\u0438",
+  "\u043F\u0440\u043E",
+  "\u0440",
+  "\u0440\u0430\u0437",
+  "\u0441",
+  "\u0441\u0430\u043C",
+  "\u0441\u0430\u043C\u0430",
+  "\u0441\u0430\u043C\u0435",
+  "\u0441\u0430\u043C\u0456",
+  "\u0441\u0432\u0456\u0439",
+  "\u0441\u0432\u043E\u0433\u043E",
+  "\u0441\u0432\u043E\u044E",
+  "\u0441\u0432\u043E\u0457",
+  "\u0441\u0435\u0431\u0435",
+  "\u0441\u0435\u0440\u0435\u0434",
+  "\u0441\u044F",
+  "\u0442",
+  "\u0442\u0430",
+  "\u0442\u0430\u043A",
+  "\u0442\u0430\u043A\u043E\u0436",
+  "\u0442\u0430\u043A\u0438\u0439",
+  "\u0442\u0430\u043A\u0438\u0445",
+  "\u0442\u0430\u043C",
+  "\u0442\u0435",
+  "\u0442\u0435\u0431\u0435",
+  "\u0442\u0435\u0436",
+  "\u0442\u0435\u043F\u0435\u0440",
+  "\u0442\u0438",
+  "\u0442\u0438\u043C",
+  "\u0442\u0438\u0445",
+  "\u0442\u043E",
+  "\u0442\u043E\u0431\u0456",
+  "\u0442\u043E\u0433\u043E",
+  "\u0442\u043E\u0434\u0456",
+  "\u0442\u043E\u043C\u0443",
+  "\u0442\u0443\u0442",
+  "\u0443",
+  "\u0443\u0432\u0435\u0441\u044C",
+  "\u0443\u0441\u0435",
+  "\u0443\u0441\u0456",
+  "\u0443\u0441\u044E",
+  "\u0443\u0441\u044F",
+  "\u0444",
+  "\u0445",
+  "\u0445\u0430\u0439",
+  "\u0445\u043E\u0447",
+  "\u0445\u043E\u0447\u0430",
+  "\u0445\u0442\u043E",
+  "\u0446",
+  "\u0446\u0435",
+  "\u0446\u0435\u0439",
+  "\u0446\u044F",
+  "\u0446\u044C\u043E\u0433\u043E",
+  "\u0446\u0438\u043C",
+  "\u0446\u0456",
+  "\u0446\u0456\u0439",
+  "\u0446\u044E",
+  "\u0447",
+  "\u0447\u0438",
+  "\u0447\u0438\u0439",
+  "\u0447\u043E\u0433\u043E",
+  "\u0447\u043E\u043C\u0443",
+  "\u0448",
+  "\u0449\u043E",
+  "\u0449\u043E\u0431",
+  "\u0449\u043E\u0441\u044C",
+  "\u044F",
+  "\u044F\u043A",
+  "\u044F\u043A\u0430",
+  "\u044F\u043A\u0435",
+  "\u044F\u043A\u0456",
+  "\u044F\u043A\u0449\u043E"
+]);
+var ukrainian = {
+  name: "ukrainian",
+  stemmer: stem,
+  stopWords,
+  tokenizer: { splitPattern: /[^a-z0-9а-яіїєґ]+/gi }
+};
+
+export { ukrainian };

package/dist/languages/yoruba.d.ts

@@ -0,0 +1,5 @@
+import { L as LanguageModule } from '../language-ClUGLuoC.js';
+
+declare const yoruba: LanguageModule;
+
+export { yoruba };

package/dist/languages/yoruba.mjs

@@ -0,0 +1,103 @@
+// src/languages/yoruba.ts
+var stopWords = /* @__PURE__ */ new Set([
+  "\xF3",
+  "n\xED",
+  "\u1E63e",
+  "r\u1EB9\u0300",
+  "t\xED",
+  "\xE0w\u1ECDn",
+  "s\xEC",
+  "s\xED",
+  "ni",
+  "n\xE1\xE0",
+  "l\xE1ti",
+  "kan",
+  "ti",
+  "\u0144",
+  "l\u1ECD",
+  "o",
+  "b\xED",
+  "pad\xE0",
+  "w\xE0",
+  "w\xE1",
+  "k\xED",
+  "p\xFAp\u1ECD\u0300",
+  "mi",
+  "w\u1ECD\u0301n",
+  "p\u1EB9\u0300l\xFA",
+  "a",
+  "\u1E63\xF9gb\u1ECD\u0301n",
+  "f\xFAn",
+  "j\u1EB9\u0301",
+  "f\u1EB9\u0301",
+  "k\xF2",
+  "j\xF9",
+  "p\xE9",
+  "\xE9",
+  "gbogbo",
+  "in\xFA",
+  "b\u1EB9\u0300r\u1EB9\u0300",
+  "j\u1EB9",
+  "\u1ECDj\u1ECD\u0301",
+  "n\xEDtor\xED",
+  "n\u01F9kan",
+  "s\xEDn\xFA",
+  "\u1E63",
+  "y\xEC\xED",
+  "\u1E63\xE9",
+  "\xE0ti",
+  "\xED",
+  "m\xE1a",
+  "n\xEDgb\xE0",
+  "mo",
+  "an",
+  "m\u1ECD\u0300",
+  "b\xE1",
+  "k\xEC",
+  "\u0144l\xE1",
+  "\u1ECD\u0300p\u1ECD\u0300l\u1ECDp\u1ECD\u0300",
+  "\u1EB9m\u1ECD\u0301",
+  "w\u1ECDn",
+  "\xF2un",
+  "l\xE8",
+  "tabi",
+  "l\xF3r\xED",
+  "n\xEDn\xFA",
+  "l\xE1b\u1EB9\u0301",
+  "l\xF3k\xE8",
+  "l\u1EB9\u0301y\xECn",
+  "n\xEDw\xE1j\xFA",
+  "l\xE1\xE0\xE1r\xEDn",
+  "n\xEDpa",
+  "g\u1EB9\u0301g\u1EB9\u0301",
+  "ir\xFA",
+  "b\u1EB9\u0301\u1EB9\u0300",
+  "ohun",
+  "\u1EB9ni",
+  "\u1EB9",
+  "k\u1ECD\u0301",
+  "w\u1ECD\u0300ny\xED",
+  "\xECw\u1ECD",
+  "\xE8mi",
+  "\xE0wa",
+  "\u1EB9yin",
+  "\u1ECD\u0300un",
+  "t\xE0b\xED",
+  "n\xEDtor\xEDp\xE9",
+  "d\xED\u1EB9\u0300",
+  "n\xECkan",
+  "p\xE0\xE1p\xE0\xE1",
+  "s\xEDb\u1EB9\u0300"
+]);
+var yoruba = {
+  name: "yoruba",
+  stemmer: null,
+  stopWords,
+  tokenizer: {
+    splitPattern: /[^a-zA-ZàáèéìíòóùúẹọṣńǹÀÁÈÉÌÍÒÓÙÚẸỌṢŃǸ0-9\p{M}]+/giu,
+    normalizeDiacritics: false,
+    minTokenLength: 1
+  }
+};
+
+export { yoruba };

package/dist/languages/zulu.d.ts

@@ -0,0 +1,5 @@
+import { L as LanguageModule } from '../language-ClUGLuoC.js';
+
+declare const zulu: LanguageModule;
+
+export { zulu };